Fwd: follow up from our talk / responce to your paper

[Scott Wheeler]

On Thursday 03 February 2005 22:40, Stanislav Karchebny wrote:
> In my original idea i proposed to store metadata in one of formats, either:
> a) <filename>.metadata file
> b) ./.metadata/<filename>
> c) ~/.metadata/<path>/<filename>
> d) filesystem EAs for <filename>
> All this depends on filesystem storage and is not too flexible. It also
> should be very slow.
>
> So, probably a combination of file content digests, metadata matches and
> filenames should serve to locate given object (i fear there will be many
> misses on a large scale - e.g. on a network of many computers files can be
> duplicate, but even files seemingly same can be different {otoh, if we find
> the same content in file B instead of file A, it is no harm unless we try
> to write to it}).

Just serving as a metadata cache would be (a) really slow and (b) not really 
useful for much, especially not to the queryable graph type stuff that's the 
essence of the klink idea.

You might want to have a look at some of the stuff in kdenonbeta/klink -- the 
DESIGN document is outdated, but should give you an idea on how some of the 
thoughts evolved.

-Scott

-- 
I mean, if 10 years from now, when you are doing something quick and dirty, 
you suddenly visualize that I am looking over your shoulders and say to 
yourself, "Dijkstra would not have liked this", well that would be enough 
immortality for me.
-E. W. Dijkstra